Does anyone have any insight into these differing output formats/model structures that could guide us? My understanding is that with a classic ANOVA & Tukey post-hoc test, a significant pairwise p-value in the Tukey test is appropriate to interpret even if the overall ANOVA is not significant but perhaps I'm wrong, or perhaps it's not an apt analogy for this situation. But it does not make sense to me that the glm() function in R would present output this way if that were the case. She is arguing that based on her JMP results, it is inappropriate for me to interpret the PLAIN treatment as being significantly different from the CONTROL treatment, and that I should report no significant effects at all of Audience Treatment on mating success.
#CORRELATION COEFFICIENT JMP HOW TO#
So for this example, JMP provided her with a non-significant p-value for 'Audience Treatment' as a whole, and won't even produce pairwise p-values for each pair of treatment groups because the overall p is insignificant (or at least, we don't know how to force JMP to do this). However, JMP formats output differently from R: for a categorical explanatory variable with more than two levels, JMP provides an 'overall' p-value rather than breaking it down into pairwise comparisons to a reference level as in R. This has worked well in our lab as a 'check' to make sure everyone is confident that we are analyzing data adequately (we're biologists, not statisticians lol). When she works with students who use R, she likes to build models with the exact same variables/covariates in JMP and compare the results to see if they are (qualitatively) the same. The problem is that my advisor works primarily in JMP. So based on this I can interpret the behavior of the PLAIN audience treatment as being significantly different from the behavior of the CONTROL audience treatment (the reference level) per the p-value of 0.0360.
I set my CONTROL treatment as the reference level for this GLM so my understanding of the output here is that each of the treatment groups listed in the output is being compared to the reference treatment group and R is producing t- and p-values for each of these pairwise comparisons. I obtained the following output: Coefficients:
I conducted an experiment in which I placed focal fish into one of four different 'audience' treatment groups (CONTROL, BLACK, PLAIN, and FEMALE) and observed their mating behaviors to see if the audience treatment explained any of the variation in mating success. I am building a generalized linear model using the stats::glm() function in R for my masters thesis.
Hello all, I'm new to reddit so I hope this is the best place to ask this question!
0 kommentar(er)
